Receptive Fields of Single Cells and Topography in Mouse Visual Cortex URSULA C . DRAGER I Mnx-Plnnck-Institutfiil- Biophysiknlische C h e m i e , 0-34Gottingeti Nikolnzisberg, Am Fnssberg, West Cemnony
ABSTRACT The visual cortex was studied in the mouse (C57 Black/GJ strain) by recording from single units, and a topographic map of the visual field was constructed. Forty-five percent of the neurons i n striate cortex responded best to oriented line stimuli moving over their receptive fields; they were classified a s simple (17%), complex (25% ) and hypercomplex (3%). Of all preferred orientations horizontal was most common. Fifty-five percent of receptive fields were circularly symmetric: these were on-center (25% ), off-center ( 7 % ) and homogeneous on-off in type (23%). Optimal stimulus velocities were much higher than those reported in the cat, mostly varying between 20" and 300"lsec. The field of vision common to the two eyes projected to more than one-third of the striate cortex. Although the contralateral eye provided the dominating influence on cells in this binocular area, more than two-thirds of cells could also be driven through the ipsilateral eye. The topography of area 1 7 was similar to that found in other mammals: the upper visual field projected posteriorly, the most nasal part mapped onto the lateral border. Here the projection did not end a t the vertical meridian passing through the animal's long axis, but proceeded for at least 10" into the ipsilateral hemifield of vision, so that a t least 20" of visual field were represented in both hemispheres. The magnification in area 17 was rather uniform throughout the visual field. In a n area lateral to area 1 7 (18a) the fields were projected in condensed mirror image fashion with respect to the arrangement of area 17. Medial to area 17 a third visual area (area 18) was again related to 1 7 a s a condensed mirror image.
Neurophysiologists have devoted little attention to the mouse and have particularly neglected its visual system. This is not only due to the difficulties in handling a small animal in a long lasting neurophysiological experiment, but rather because such a study does not fit with two of the most widely accepted practices in neurophysiology: either to investigate a very simple animal having only a small number of identifiable neurons, with the prospect of understanding its whole nervous system in terms of the interaction of all its components, or to study one subsystem, such as the visual pathway, in an animal in which that system is developed to a very refined stage. The mouse is neither a simple animal nor is its visual capability outstanding; on the contrary there is no doubt from behavioral studies that its visual sense is not the dominating one and indeed is rather coarse (for references, J. COMP. NEUR.,160. 269-290
see Fuller and Wimer, '66). One argument for a study of the mouse visual system stems from an interest in comparative physiology, given the extensive studies already available in the cat, monkey and also the more closely related rabbit. But the strongest motivation for a functional analysis of the mouse central nervous system is the availability of genetically pure strains and of many well-identified neurological mutants (Sidman et al., '65). Since the visual pathway represents perhaps the most thoroughly investigated part of the mammalian central nervous system, it seemed easiest to take advantage of existing knowledge and to work out a part of the mouse visual pathway as a basis for future studies of mutant mice. For such a study it does not primarily matter I Present address: Harvard Medical School, Department of Neurobiology, 25 Shattuck Street. Boston, Massachusetts 02115.
269
2 70
URSULA C DRAGER
dampen pulsations the best method was to cement the skin to a ring cut from a silicone tube using a water-binding glue (Cyanolit), and to fill the well so formed METHODS with agar. To keep the optics transparent in mice Included in the analysis of this paper are recordings from 56 mice of the C57 presented a major problem. Various anesBlack/6j strain obtained from the Zentral- thetics, cold, anoxia, or other stress can institut fur Versuchstierzucht in Hannover, lead to a reversible cataract in the anteand two mice that were a cross between rior capsule of the lens. Fraunfelder and the local wild mouse and several laboratory Burns ('70) found that the osmotic presstrains. The mice were about 4 to 8 months sure in the anterior chamber increases old and their weights ranged between 20 and recommended using contact lenses. and 26 grams. There was no obvious dif- In the first experiments the eyes were proference between the two strains in the tected with silicone fluid (1,000 sc. Dow Corning); later contact lenses of a variety results to be reported. The mouse was anesthetized with an of sizes were used. Most animals were best initial dose of 60 mg/kg of pentobarbital fitted with lenses with a radius of curvaintraperitoneally; this was supplemented ture between 1.50 and 1.55 mm. The later, as required, by additional 0.02 mg mouse eye a s measured with an ophthal15 doses. Atropine (0.03 ml of a 1 % solution) moscope is highly hyperopic, about was used to counteract vagotonic effects diopters. This is probably an error due to of the anesthetic. To prevent cerebral the distance between the light reflecting edema, which is easily caused merely by layer and the receptor layer (Bruckner, mechanical vibrations from drilling the '51; Glickstein and Millodot, '70), a disskull, about 1 mg of prednisolon was in- tance that remains fairly constant among jected intramuscularly. Later a single dose different species, but becomes increasingly of 0.12 mg of chlorprothixene (Truxal) important with decreasing eye size. Here was given intramuscularly; this is a tran- it was presumed that the mouse is emmequilizer whose effect is synergistic with tropic, and contact lenses were selected so that of pentobarbital. With this combina- a s not to change the appearance of the tion at optimal anesthetic levels for re- fundus when viewed with a n ophthalmocording from single cells, the mouse sits scope. No attempt was made to correct for quietly without moving but reacts vigor- focus on the screen, since the depth of ously if pinched. The anesthetic level must focus in such a small eye was assumed be light if cells are to respond well. Even to be large; with ophthalmometric methods when no artificial respiration was used such a correction would have been most the trachea was always cannulated, since difficult. The mouse was placed 27 cm from a otherwise i t tended to become clogged with salivary secretions. Rectal temperature 1 X 1.4 meter translucent tangent screen. was maintained at 36.5"C by a manually The long axis of the mouse formed a n controlled heating pad. The head was sup- angle of 65 degrees with the screen; this ported by a holder that was clamped to angle was chosen as a compromise, given an LPC stereotaxic system for cats. The the laterally positioned eyes and a possihorizontal plane of this holder lies between bly larger central representation of the the intraaural line and a point 2 mm anterior field of vision. The arrangement above the incisor bar and provides the was kept constant in all experiments. The same inclination as that used by Monte- screen covered a rather large sector of the murro and Dukelow ('72) in their stereo- field of vision through one eye (120 X 130 taxic atlas. The skull over the visual cortex degrees), but the peripheral parts were on one side was removed cautiously by projected in a somewhat distorted manfirst drilling away the external and spongy ner. As a retinal landmark the optic disc layers of bone and then tearing off the was projected on the screen with an ophinternal layer with forceps. Great care thalmoscope. It projected roughly 60 dewas taken to leave the dura intact. To pro- grees lateral to the animal's midline and tect the cortex against drying and to 40 degrees above the horizontal. No cenhow well the visual sense is developed or how important it may be for the mouse, as long as an order of any kind is revealed.
+
MOUSE V I S U A L CORTEX
271
tral retinal area could be determined oph- several small receptive fields were observed for up to three hours; during this time thalmoscopically. Most recordings were performed with large rapid eye movements were never laquer coated tungsten electrodes. In a seen and slow drifts never exceeded 2-3" few experiments glass pipettes were used; per hour, a negligible error compared to these were filled with 1.5 molar potassium the average size of the receptive fields. Many preliminary experiments were citrate and had a resistance of 3-6 megohms. In several experiments with tung- nevertheless performed to achieve relaxasten electrodes the recording tracks were tion by curarizing with Flaxedil and resmarked by electrolytic lesions made by pirating artificially by means of a rodent passing 3 PA for 2 seconds through the respirator. It was found that relaxation electrode tip (DC, electrode negative). Ani- was possible but rather tedious, since the mals were perfused at the end of the ex- cortex tended to become unresponsive or periment and their brains prepared for give epileptiform discharges unless the histology. Impulses from single cells were respiration volume was very carefully regrecorded and amplified by conventional ulated. The few good units recorded in recording methods, displayed on an oscil- respirated animals were similar to the loscope, monitored by a loudspeaker, and ones recorded without relaxation and are sometimes stored on an Ampex tape re- included in the results without special refcorder. For immediate analysis a two erence. As residual eye movements and the continuous administration of aneschannel electronic counter was used. Various stimuli were projected manu- thetics did not to any serious degree interally on the screen by a Zeiss handlamp or fere with the study most experiments were a slide projector. Automatically generated done without relaxation. stimuli consisted of circles of various diRESULTS ameters flashed by means of an electroFour hundred units and some unit clusmagnetic shutter driven by a Hivotronic impulse generator. To generate moving ters were recorded in area 17 and the distimuli a servo-driven oscillating projector rectly adjoining visual cortex of 57 mice. system was used, in which different stim- Many units in the early experiments and uli could be moved back and forth over some in the later experiments were not the screen with variable speed and ampli- completely investigated. For some the only tude and rotated in any direction. The information obtained was the location of screen was indirectly illuminated from a the receptive field on the screen, and in distance by two tungsten bulbs; back- some the preferred direction of movement ground illuminance was about 0.5 cd/mz was determined, but too little information and stimulus intensity ranged between 2 was obtained to permit further classification. Not all cells were tested with both and 10 cd/m'. In several mice direct measurements eyes. In experiments in which glass elecwere made to determine whether eye move- trodes were used there were sometimes ments were a serious problem in the lightly doubts concerning the identification of the anesthetized animals. For that purpose a point of contact with the cortical surface tiny mirror (about 1/2 mm?) was glued to and consequently the depth of a recorded one eye with water-binding cement, and cell. The samples of neurons tested for difa narrow beam of light was reflected from ferent characteristics thus vary slightly this mirror to the screen. Under these con- in number. For construction of the topographic map ditions there was a continuous vibration of about 0.25" to 0.5" synchronous with unit clusters as well as isolated units were respiratory excursions, but no large rapid used. For all other purposes all units had eye movements were visible. In some mice to be well enough isolated to permit disthere was a slow continuous drift in eye tinction between fiber and cell spikes position of up to 15" in three hours. This (Hubel, '60); of these only cells were anamay have been partly an artifact due to lyzed and classified. Additional evidence the irritation by the glue and the weight that the cell classes to be reported repreof the mirrow compared to that of the sent cortical cells and not fibers is the small eye. In actual recording experiments observation that some of the units in all
2 72
URSULA C. DRAGER
classes were binocular or showed prolonged injury discharges. The following detailed analysis of single units applies only to area 17. Too few units were identified in areas 18 and 18a to provide more than a general impression of the population in these areas. Units in mouse area 17, like those in cat and monkey, reacted poorly to changes in overall illumination. Moving patterns of various shapes and contrast were usually the most effective stimuli, but for purposes of mapping separate field subdivisions stationary stimuli were commonly flashed on ( ) and off ( - ). Two main groups of cortical cells were found. Cells of one group responded equally well to all directions of movement and
+
were not concerned with stimulus orientation. A second smaller group of cells strongly preferred a properly oriented elongated stimulus swept across the field in one or both of the two opposite directions. A. Cells with non-oriented receptive fields In this group three subtypes were found: on-center cells, off-center cells, and a heterogenous group of onaff cells which are here called “fast cells” because of their common characteristic of reacting to rapid movement.
&-center cells On-center cells constituted 25% of all classified cells in mouse area 17. In figure +
OFF-CENTER CELLS
ON-CENTER CELLS
0 _ -
+
-
+
FAST CELLS
SIMPLE CELLS -
- _ + +++
--p@ + + ++++
-
+++y=
0
COMPLEX CELLS
+++ + ++
HYPERCOMPLEX CELLS 00
Fig. 1
560
Maps of the different receptive field types found in mouse area 17, drawn to scale.
273
M O U S E VISUAL CORTEX
1 a few on-center fields are drawn; the were rather difficult to drive, giving only average diameter of their center plus sur- a few spikes even to the most effective round was 35" and ranged from 18" to stimulus. A spot of light flashed into the 5 0 " ; the mean diameter of field centers center evoked a transient off-response; a was 15", ranging from 6 " to 35". Oncen- stimulus flashed within the surround proter cells reacted with maximum response duced a weak on-discharge. The best way to a light spot flashed in the field center to stimulate these cells was to move a light and filling it. Responses to flashed stimuli stimulus outwards from the center into were transient, and this was found for all the surround with a velocity of about 50" cells in mouse area 17. The power of the to 15O"lsec. surround in suppressing the center response was very striking: in contrast e.g. "Fast cells" (on-off cells) to on-center cells in cat lateral geniculate This group of cells comprised 23% of these cells reacted poorly to a very long the total sample of classified cortical cells. slit moved over the whole field and re- Their common characteristic was a homogsponded either weakly or not at all to enous receptive field without any obvious changes in diffuse illumination. Despite subareas but with a substructure with this strong influence of the surround over complex properties. The average field size the center, it was not easy to obtain an was 22", ranging from 12" to 4 5 " . These off-response from the surround with flashed cells were easy to find mainly because of stimuli. A carefully placed annulus of their characteristic spontaneous highlight flashed in the surround usually frequency bursting firing pattern, which evoked only a weak off-discharge, and for in some cells reached levels as high as 15 determining the extent of the surround spikes per second. The best stimulus was the best method was to produce a large an edge or restricted spot of any shape spot of light with an iris diaphragm, and moved very rapidly over the receptive field to close the diaphragm abruptly in small at rates of 100-1,000"/sec. Many of these steps. The off-surrounds in figure 1 were cells also reacted to slow movement (in the obtained by this method. A spot of light range of 20" to 5 O " / s e c . ) , but they then smaller than the field center moved from responded with a series of high-frequency the surround into the center was also usu- bursts that could easily be confused with ally a much more effective stimulus than their spontaneous activity. The stimulus a flash. The optimal speed for such a spot form and size were not critical; for optimal was about 30" to 150"Isec and for a given response a slit or edge had to cover the cell the speed was often quite critical. On- entire field, but it could also be much center cells were easy to find since they longer. When a light square was used most had a high spontaneous activity (about 50 cells reacted preferentially to the leading to 150 spikes per minute) and generally edge and less to the trailing edge. A reresponded vigorously. stricted light stimulus was usually better than a dark one, but some cells were Offcenter cells highly sensitive even to the faintest movOff-center cells were rare; only 15 were ing shadow. When tested with a moving found in area 17 (7% of the total sample). grid some of the fast cells could resolve a Their receptive fields were very large, with spacing finer than the field size; for exan average center diameter of 51 ', rang- ample, one cell with a field 12" in diaming from 15" to 80". Spontaneous activity eter could resolve a grid frequency of 6". was low (5-10 spikeslmin), and the cells A s with other cortical cells, fast cells reTABLE 1
Frequency of 224 cells in nren 17 ~
Simple cells
Complex cells
Hypercomplex cells
On-center cells
Off-center cells
Fast cells
38
56
6
56
15
53
= 17%
=25%
=3%
=25%
=7%
=234
274
U R S U L A C DRAGER
A
B SIMPLE
+
SIMPLE CELL ( 2 9 2 ) 10 c y c l e s
w i t h 4'120'
speed 15'/sec
+
+
C E L L 1295)
r e c e p l ~ v ef i e l d StI"c1Yr.:
+
Ill1 t
i
d o w n : light
+
3-
edge
. 10"
UP: d a r k
200
r o c y c I e s , light square 25Ox 2 5 4 speed
nasal.
' I '
down
down
Figure 2 A
Figure 2 B
sponded poorly or not at all to changes in background illumination. Flashing a light spot evoked a transient on-off response all over the receptive field. Usually the oncomponent was stronger, but in cells preferring a moving dark stimulus the offresponse dominated. The on-off structure was hard to demonstrate in some cells and easy in others. When it was easy the cells could be triggered by flickering a light spot within their field, some following a flicker frequency of up to 14/sec. Usually fast cells had no surround. In some cells the field borders appeared indistinct, attenuating gradually in the periphery. When tested by closing or opening a n iris diaphragm, a few cells had a weak surround of either on-, off-, or onfoff-type, and a few cells seemed to be intermediate in form between this group and on-center cells.
These categories appeared to be similar to those described in visual cortex of other mammals.
B. Cells with oriented receptive fields Cells in this group fell into three subtypes; simple, complex and hypercomplex.
Simple cells Thirty-eight cells of the 224 included in table 1, or 17% of all classified cells in area 17, were categorized as simple. By flashing slits or squares of light the receptive fields could be shown to consist of an excitatory area flanked by inhibitory areas on one or both sides, or more rarely an inhibitory area flanked by two excitatory areas. Unless the flashing stimulus was exactly placed to fill a whole area, most cells did not respond at all. Usually it was more difficult to evoke a n off-response from inhibitory flanks than to demonstrate excitatory areas, but the existence of inhibitory flanks could be demonstrated indirectly, since flashing light on both areas simultaneously abolished the on-response from excitatory areas. As in all other cortical cell types the reaction to changes in
2 75
MOUSE V I S U A L CORTEX
C SIMPLE CELL
(368)
10 c y c l e r with 5*r 4 0 ' weed 40"/stc:
silt
k down
Figure 2C Fig. 2 Responses of three simple cells to moving line stimuli in various orientations. For each cell receptive field m a p is shown at the top. Responses are plotted in polar coordinates. For each point plotted, distance from origin represents number of spikesilo sweeps, direction from origin represents direction of movement. A. Cell responding best to a n upward moving slit. Diagram appears somewhat distorted since field projected on the periphery of the tangent screen. B. Cell responding best to edge with dark below. C. Cell responding to slit and dark bar; for this cell optimum direction of movement reversed with contrast reversal.
diffuse illumination was generally inconsistent or rather poor in simple cells. When there was a reaction it was dominated by the strongest subarea: the cells tended to give a weak on-response to diffuse light if the optimal stimulus was a slit, and a weak off-reation when it was a bar. Figure 1 shows a representative sample of all receptive-field types found in area 17, drawn to the same scale. The mean diameter of the total receptive fields of simple cells was 23.5", ranging from 6 " to 55". The mean width of the main area was 10 ranging from 2 to 30 O. The extent of the field along the orientation axis was usually shorter than along the line perpendicular to this (the movement direction), particularly in large fields. It was sometimes difficult to determine the exact extent of the field along the axis, because a stimulus was only effective when it virtually filled a field subdivision. Similarly the expanse of the side flanks in a direction perpendicular to the orientation O ,
O
might have been larger in some fields than could be demonstrated. The most effective stimulus was a slit, or less frequently a dark bar or a l i g h t d a r k edge, moved at right angles to the orientation of the subareas of the receptive field. To evoke a n optimal response the stimulus had to be long enough to cover the entire length of the field, but it could be much longer without loss of effectiveness. The optimal speed of motion ranged from 5 to about 200 degrees per second and was critical for each cell. The thickness of a bar or slit, on the other hand, was not very critical: optimally it was about one-half to two-thirds of the width of the main area. To document the cells' reactions to oriented movement, diagrams like those in figure 2 were made by sweeping a line, bar or edge over the receptive field and counting the number of spikes for each direction: the curves of response (in impulses per 10 sweeps) versus movement direction are plotted in polar coordinates. In some
276
URSULA C. DRAGER
curves (e.g., figs. 2A, 3A,C) the two preferred directions are not exactly 180" apart. This is probably a distortion related to the eccentric position of the receptive fields on the tangent screen. Simple cells usually had a low spontaneous activity, at least in the anesthetic state used in these experiments and compared to other cell types; they usually generated 2 to 10 spikes per minute spontaneously and only responded with a few spikes even to the optimal stimulus. This sluggish behavior made the detailed mapping of some fields difficult. The inhibitory flanks were often especially hard to demonstrate with a stationary stimulus, but their presence could usually be shown indirectly by comparing the response to a slit confined to the excitatory region with the response to a larger slit that included some or all of the inhibitory regions. In those cells in which separate "on" and "off" subareas could clearly be demonstrated, the orientation of a moving line and the most effective direction of back and forth movement could very often be predicted from the arrangement of subareas. Some cells with a side by side arrangement of two antagonistic areas responded only to a simple contrast moved in either direction. One example is the cell in figure 2B: this cell responded only when a light edge was moved from above into the on-area or when a dark edge was moved from below into the off-area - that is, it reacted to moving a light-above-dark contrast in both directions. Of cells responding specifically to edges, including simple and complex cells, the great majority preferred horizontal edges with dark below. Slits were very ineffective in activating such cells. Most cells were much more sensitive to a slit than to a dark bar; those preferring dark stimuli usually had one or two exceptionally strong inhibitory areas in their fields. For any given cell the preferred direction of movement was generally the same for a dark bar a s for a slit. The cell Fig. 3 Polar plots of responses of four complex cells. A. Cell responding best to a moving edge. B. Complex cell whose field could be subdivided, by flashing stationary stimuli, into a n upper area which gave "on" responses and a lower area giving "off' responses. T h e tuning curve of this cell was the narrowest recorded in this series. C. Cell responding best to a dark bar. D. Cell responding best to a narrow slit of light.
A
COMPLEX
CELL (235)
10 c y c l e s w i t h 5 0 % 50' square speed 30"//nac
TO'
down lighl edge
UP: dark edge
nasal c
lsmporal
' 1 ' down
Figure 3A
B COMPLEX CELL ( 2 8 5 )
/ Y
10
down
Figure 3B
MOUSE V I S U A L CORTEX
c
in figure 2C was the only exception; here a bar and a slit had diametrically opposite optimal movement directions.
COMPLEX CELL (289) IDCISI.. W)h .W.d IO.,..
ABSTRACT The visual cortex was studied in the mouse (C57 Black/GJ strain) by recording from single units, and a topographic map of the visual field was constructed. Forty-five percent of the neurons i n striate cortex responded best to oriented line stimuli moving over their receptive fields; they were classified a s simple (17%), complex (25% ) and hypercomplex (3%). Of all preferred orientations horizontal was most common. Fifty-five percent of receptive fields were circularly symmetric: these were on-center (25% ), off-center ( 7 % ) and homogeneous on-off in type (23%). Optimal stimulus velocities were much higher than those reported in the cat, mostly varying between 20" and 300"lsec. The field of vision common to the two eyes projected to more than one-third of the striate cortex. Although the contralateral eye provided the dominating influence on cells in this binocular area, more than two-thirds of cells could also be driven through the ipsilateral eye. The topography of area 1 7 was similar to that found in other mammals: the upper visual field projected posteriorly, the most nasal part mapped onto the lateral border. Here the projection did not end a t the vertical meridian passing through the animal's long axis, but proceeded for at least 10" into the ipsilateral hemifield of vision, so that a t least 20" of visual field were represented in both hemispheres. The magnification in area 17 was rather uniform throughout the visual field. In a n area lateral to area 1 7 (18a) the fields were projected in condensed mirror image fashion with respect to the arrangement of area 17. Medial to area 17 a third visual area (area 18) was again related to 1 7 a s a condensed mirror image.
Neurophysiologists have devoted little attention to the mouse and have particularly neglected its visual system. This is not only due to the difficulties in handling a small animal in a long lasting neurophysiological experiment, but rather because such a study does not fit with two of the most widely accepted practices in neurophysiology: either to investigate a very simple animal having only a small number of identifiable neurons, with the prospect of understanding its whole nervous system in terms of the interaction of all its components, or to study one subsystem, such as the visual pathway, in an animal in which that system is developed to a very refined stage. The mouse is neither a simple animal nor is its visual capability outstanding; on the contrary there is no doubt from behavioral studies that its visual sense is not the dominating one and indeed is rather coarse (for references, J. COMP. NEUR.,160. 269-290
see Fuller and Wimer, '66). One argument for a study of the mouse visual system stems from an interest in comparative physiology, given the extensive studies already available in the cat, monkey and also the more closely related rabbit. But the strongest motivation for a functional analysis of the mouse central nervous system is the availability of genetically pure strains and of many well-identified neurological mutants (Sidman et al., '65). Since the visual pathway represents perhaps the most thoroughly investigated part of the mammalian central nervous system, it seemed easiest to take advantage of existing knowledge and to work out a part of the mouse visual pathway as a basis for future studies of mutant mice. For such a study it does not primarily matter I Present address: Harvard Medical School, Department of Neurobiology, 25 Shattuck Street. Boston, Massachusetts 02115.
269
2 70
URSULA C DRAGER
dampen pulsations the best method was to cement the skin to a ring cut from a silicone tube using a water-binding glue (Cyanolit), and to fill the well so formed METHODS with agar. To keep the optics transparent in mice Included in the analysis of this paper are recordings from 56 mice of the C57 presented a major problem. Various anesBlack/6j strain obtained from the Zentral- thetics, cold, anoxia, or other stress can institut fur Versuchstierzucht in Hannover, lead to a reversible cataract in the anteand two mice that were a cross between rior capsule of the lens. Fraunfelder and the local wild mouse and several laboratory Burns ('70) found that the osmotic presstrains. The mice were about 4 to 8 months sure in the anterior chamber increases old and their weights ranged between 20 and recommended using contact lenses. and 26 grams. There was no obvious dif- In the first experiments the eyes were proference between the two strains in the tected with silicone fluid (1,000 sc. Dow Corning); later contact lenses of a variety results to be reported. The mouse was anesthetized with an of sizes were used. Most animals were best initial dose of 60 mg/kg of pentobarbital fitted with lenses with a radius of curvaintraperitoneally; this was supplemented ture between 1.50 and 1.55 mm. The later, as required, by additional 0.02 mg mouse eye a s measured with an ophthal15 doses. Atropine (0.03 ml of a 1 % solution) moscope is highly hyperopic, about was used to counteract vagotonic effects diopters. This is probably an error due to of the anesthetic. To prevent cerebral the distance between the light reflecting edema, which is easily caused merely by layer and the receptor layer (Bruckner, mechanical vibrations from drilling the '51; Glickstein and Millodot, '70), a disskull, about 1 mg of prednisolon was in- tance that remains fairly constant among jected intramuscularly. Later a single dose different species, but becomes increasingly of 0.12 mg of chlorprothixene (Truxal) important with decreasing eye size. Here was given intramuscularly; this is a tran- it was presumed that the mouse is emmequilizer whose effect is synergistic with tropic, and contact lenses were selected so that of pentobarbital. With this combina- a s not to change the appearance of the tion at optimal anesthetic levels for re- fundus when viewed with a n ophthalmocording from single cells, the mouse sits scope. No attempt was made to correct for quietly without moving but reacts vigor- focus on the screen, since the depth of ously if pinched. The anesthetic level must focus in such a small eye was assumed be light if cells are to respond well. Even to be large; with ophthalmometric methods when no artificial respiration was used such a correction would have been most the trachea was always cannulated, since difficult. The mouse was placed 27 cm from a otherwise i t tended to become clogged with salivary secretions. Rectal temperature 1 X 1.4 meter translucent tangent screen. was maintained at 36.5"C by a manually The long axis of the mouse formed a n controlled heating pad. The head was sup- angle of 65 degrees with the screen; this ported by a holder that was clamped to angle was chosen as a compromise, given an LPC stereotaxic system for cats. The the laterally positioned eyes and a possihorizontal plane of this holder lies between bly larger central representation of the the intraaural line and a point 2 mm anterior field of vision. The arrangement above the incisor bar and provides the was kept constant in all experiments. The same inclination as that used by Monte- screen covered a rather large sector of the murro and Dukelow ('72) in their stereo- field of vision through one eye (120 X 130 taxic atlas. The skull over the visual cortex degrees), but the peripheral parts were on one side was removed cautiously by projected in a somewhat distorted manfirst drilling away the external and spongy ner. As a retinal landmark the optic disc layers of bone and then tearing off the was projected on the screen with an ophinternal layer with forceps. Great care thalmoscope. It projected roughly 60 dewas taken to leave the dura intact. To pro- grees lateral to the animal's midline and tect the cortex against drying and to 40 degrees above the horizontal. No cenhow well the visual sense is developed or how important it may be for the mouse, as long as an order of any kind is revealed.
+
MOUSE V I S U A L CORTEX
271
tral retinal area could be determined oph- several small receptive fields were observed for up to three hours; during this time thalmoscopically. Most recordings were performed with large rapid eye movements were never laquer coated tungsten electrodes. In a seen and slow drifts never exceeded 2-3" few experiments glass pipettes were used; per hour, a negligible error compared to these were filled with 1.5 molar potassium the average size of the receptive fields. Many preliminary experiments were citrate and had a resistance of 3-6 megohms. In several experiments with tung- nevertheless performed to achieve relaxasten electrodes the recording tracks were tion by curarizing with Flaxedil and resmarked by electrolytic lesions made by pirating artificially by means of a rodent passing 3 PA for 2 seconds through the respirator. It was found that relaxation electrode tip (DC, electrode negative). Ani- was possible but rather tedious, since the mals were perfused at the end of the ex- cortex tended to become unresponsive or periment and their brains prepared for give epileptiform discharges unless the histology. Impulses from single cells were respiration volume was very carefully regrecorded and amplified by conventional ulated. The few good units recorded in recording methods, displayed on an oscil- respirated animals were similar to the loscope, monitored by a loudspeaker, and ones recorded without relaxation and are sometimes stored on an Ampex tape re- included in the results without special refcorder. For immediate analysis a two erence. As residual eye movements and the continuous administration of aneschannel electronic counter was used. Various stimuli were projected manu- thetics did not to any serious degree interally on the screen by a Zeiss handlamp or fere with the study most experiments were a slide projector. Automatically generated done without relaxation. stimuli consisted of circles of various diRESULTS ameters flashed by means of an electroFour hundred units and some unit clusmagnetic shutter driven by a Hivotronic impulse generator. To generate moving ters were recorded in area 17 and the distimuli a servo-driven oscillating projector rectly adjoining visual cortex of 57 mice. system was used, in which different stim- Many units in the early experiments and uli could be moved back and forth over some in the later experiments were not the screen with variable speed and ampli- completely investigated. For some the only tude and rotated in any direction. The information obtained was the location of screen was indirectly illuminated from a the receptive field on the screen, and in distance by two tungsten bulbs; back- some the preferred direction of movement ground illuminance was about 0.5 cd/mz was determined, but too little information and stimulus intensity ranged between 2 was obtained to permit further classification. Not all cells were tested with both and 10 cd/m'. In several mice direct measurements eyes. In experiments in which glass elecwere made to determine whether eye move- trodes were used there were sometimes ments were a serious problem in the lightly doubts concerning the identification of the anesthetized animals. For that purpose a point of contact with the cortical surface tiny mirror (about 1/2 mm?) was glued to and consequently the depth of a recorded one eye with water-binding cement, and cell. The samples of neurons tested for difa narrow beam of light was reflected from ferent characteristics thus vary slightly this mirror to the screen. Under these con- in number. For construction of the topographic map ditions there was a continuous vibration of about 0.25" to 0.5" synchronous with unit clusters as well as isolated units were respiratory excursions, but no large rapid used. For all other purposes all units had eye movements were visible. In some mice to be well enough isolated to permit disthere was a slow continuous drift in eye tinction between fiber and cell spikes position of up to 15" in three hours. This (Hubel, '60); of these only cells were anamay have been partly an artifact due to lyzed and classified. Additional evidence the irritation by the glue and the weight that the cell classes to be reported repreof the mirrow compared to that of the sent cortical cells and not fibers is the small eye. In actual recording experiments observation that some of the units in all
2 72
URSULA C. DRAGER
classes were binocular or showed prolonged injury discharges. The following detailed analysis of single units applies only to area 17. Too few units were identified in areas 18 and 18a to provide more than a general impression of the population in these areas. Units in mouse area 17, like those in cat and monkey, reacted poorly to changes in overall illumination. Moving patterns of various shapes and contrast were usually the most effective stimuli, but for purposes of mapping separate field subdivisions stationary stimuli were commonly flashed on ( ) and off ( - ). Two main groups of cortical cells were found. Cells of one group responded equally well to all directions of movement and
+
were not concerned with stimulus orientation. A second smaller group of cells strongly preferred a properly oriented elongated stimulus swept across the field in one or both of the two opposite directions. A. Cells with non-oriented receptive fields In this group three subtypes were found: on-center cells, off-center cells, and a heterogenous group of onaff cells which are here called “fast cells” because of their common characteristic of reacting to rapid movement.
&-center cells On-center cells constituted 25% of all classified cells in mouse area 17. In figure +
OFF-CENTER CELLS
ON-CENTER CELLS
0 _ -
+
-
+
FAST CELLS
SIMPLE CELLS -
- _ + +++
--p@ + + ++++
-
+++y=
0
COMPLEX CELLS
+++ + ++
HYPERCOMPLEX CELLS 00
Fig. 1
560
Maps of the different receptive field types found in mouse area 17, drawn to scale.
273
M O U S E VISUAL CORTEX
1 a few on-center fields are drawn; the were rather difficult to drive, giving only average diameter of their center plus sur- a few spikes even to the most effective round was 35" and ranged from 18" to stimulus. A spot of light flashed into the 5 0 " ; the mean diameter of field centers center evoked a transient off-response; a was 15", ranging from 6 " to 35". Oncen- stimulus flashed within the surround proter cells reacted with maximum response duced a weak on-discharge. The best way to a light spot flashed in the field center to stimulate these cells was to move a light and filling it. Responses to flashed stimuli stimulus outwards from the center into were transient, and this was found for all the surround with a velocity of about 50" cells in mouse area 17. The power of the to 15O"lsec. surround in suppressing the center response was very striking: in contrast e.g. "Fast cells" (on-off cells) to on-center cells in cat lateral geniculate This group of cells comprised 23% of these cells reacted poorly to a very long the total sample of classified cortical cells. slit moved over the whole field and re- Their common characteristic was a homogsponded either weakly or not at all to enous receptive field without any obvious changes in diffuse illumination. Despite subareas but with a substructure with this strong influence of the surround over complex properties. The average field size the center, it was not easy to obtain an was 22", ranging from 12" to 4 5 " . These off-response from the surround with flashed cells were easy to find mainly because of stimuli. A carefully placed annulus of their characteristic spontaneous highlight flashed in the surround usually frequency bursting firing pattern, which evoked only a weak off-discharge, and for in some cells reached levels as high as 15 determining the extent of the surround spikes per second. The best stimulus was the best method was to produce a large an edge or restricted spot of any shape spot of light with an iris diaphragm, and moved very rapidly over the receptive field to close the diaphragm abruptly in small at rates of 100-1,000"/sec. Many of these steps. The off-surrounds in figure 1 were cells also reacted to slow movement (in the obtained by this method. A spot of light range of 20" to 5 O " / s e c . ) , but they then smaller than the field center moved from responded with a series of high-frequency the surround into the center was also usu- bursts that could easily be confused with ally a much more effective stimulus than their spontaneous activity. The stimulus a flash. The optimal speed for such a spot form and size were not critical; for optimal was about 30" to 150"Isec and for a given response a slit or edge had to cover the cell the speed was often quite critical. On- entire field, but it could also be much center cells were easy to find since they longer. When a light square was used most had a high spontaneous activity (about 50 cells reacted preferentially to the leading to 150 spikes per minute) and generally edge and less to the trailing edge. A reresponded vigorously. stricted light stimulus was usually better than a dark one, but some cells were Offcenter cells highly sensitive even to the faintest movOff-center cells were rare; only 15 were ing shadow. When tested with a moving found in area 17 (7% of the total sample). grid some of the fast cells could resolve a Their receptive fields were very large, with spacing finer than the field size; for exan average center diameter of 51 ', rang- ample, one cell with a field 12" in diaming from 15" to 80". Spontaneous activity eter could resolve a grid frequency of 6". was low (5-10 spikeslmin), and the cells A s with other cortical cells, fast cells reTABLE 1
Frequency of 224 cells in nren 17 ~
Simple cells
Complex cells
Hypercomplex cells
On-center cells
Off-center cells
Fast cells
38
56
6
56
15
53
= 17%
=25%
=3%
=25%
=7%
=234
274
U R S U L A C DRAGER
A
B SIMPLE
+
SIMPLE CELL ( 2 9 2 ) 10 c y c l e s
w i t h 4'120'
speed 15'/sec
+
+
C E L L 1295)
r e c e p l ~ v ef i e l d StI"c1Yr.:
+
Ill1 t
i
d o w n : light
+
3-
edge
. 10"
UP: d a r k
200
r o c y c I e s , light square 25Ox 2 5 4 speed
nasal.
' I '
down
down
Figure 2 A
Figure 2 B
sponded poorly or not at all to changes in background illumination. Flashing a light spot evoked a transient on-off response all over the receptive field. Usually the oncomponent was stronger, but in cells preferring a moving dark stimulus the offresponse dominated. The on-off structure was hard to demonstrate in some cells and easy in others. When it was easy the cells could be triggered by flickering a light spot within their field, some following a flicker frequency of up to 14/sec. Usually fast cells had no surround. In some cells the field borders appeared indistinct, attenuating gradually in the periphery. When tested by closing or opening a n iris diaphragm, a few cells had a weak surround of either on-, off-, or onfoff-type, and a few cells seemed to be intermediate in form between this group and on-center cells.
These categories appeared to be similar to those described in visual cortex of other mammals.
B. Cells with oriented receptive fields Cells in this group fell into three subtypes; simple, complex and hypercomplex.
Simple cells Thirty-eight cells of the 224 included in table 1, or 17% of all classified cells in area 17, were categorized as simple. By flashing slits or squares of light the receptive fields could be shown to consist of an excitatory area flanked by inhibitory areas on one or both sides, or more rarely an inhibitory area flanked by two excitatory areas. Unless the flashing stimulus was exactly placed to fill a whole area, most cells did not respond at all. Usually it was more difficult to evoke a n off-response from inhibitory flanks than to demonstrate excitatory areas, but the existence of inhibitory flanks could be demonstrated indirectly, since flashing light on both areas simultaneously abolished the on-response from excitatory areas. As in all other cortical cell types the reaction to changes in
2 75
MOUSE V I S U A L CORTEX
C SIMPLE CELL
(368)
10 c y c l e r with 5*r 4 0 ' weed 40"/stc:
silt
k down
Figure 2C Fig. 2 Responses of three simple cells to moving line stimuli in various orientations. For each cell receptive field m a p is shown at the top. Responses are plotted in polar coordinates. For each point plotted, distance from origin represents number of spikesilo sweeps, direction from origin represents direction of movement. A. Cell responding best to a n upward moving slit. Diagram appears somewhat distorted since field projected on the periphery of the tangent screen. B. Cell responding best to edge with dark below. C. Cell responding to slit and dark bar; for this cell optimum direction of movement reversed with contrast reversal.
diffuse illumination was generally inconsistent or rather poor in simple cells. When there was a reaction it was dominated by the strongest subarea: the cells tended to give a weak on-response to diffuse light if the optimal stimulus was a slit, and a weak off-reation when it was a bar. Figure 1 shows a representative sample of all receptive-field types found in area 17, drawn to the same scale. The mean diameter of the total receptive fields of simple cells was 23.5", ranging from 6 " to 55". The mean width of the main area was 10 ranging from 2 to 30 O. The extent of the field along the orientation axis was usually shorter than along the line perpendicular to this (the movement direction), particularly in large fields. It was sometimes difficult to determine the exact extent of the field along the axis, because a stimulus was only effective when it virtually filled a field subdivision. Similarly the expanse of the side flanks in a direction perpendicular to the orientation O ,
O
might have been larger in some fields than could be demonstrated. The most effective stimulus was a slit, or less frequently a dark bar or a l i g h t d a r k edge, moved at right angles to the orientation of the subareas of the receptive field. To evoke a n optimal response the stimulus had to be long enough to cover the entire length of the field, but it could be much longer without loss of effectiveness. The optimal speed of motion ranged from 5 to about 200 degrees per second and was critical for each cell. The thickness of a bar or slit, on the other hand, was not very critical: optimally it was about one-half to two-thirds of the width of the main area. To document the cells' reactions to oriented movement, diagrams like those in figure 2 were made by sweeping a line, bar or edge over the receptive field and counting the number of spikes for each direction: the curves of response (in impulses per 10 sweeps) versus movement direction are plotted in polar coordinates. In some
276
URSULA C. DRAGER
curves (e.g., figs. 2A, 3A,C) the two preferred directions are not exactly 180" apart. This is probably a distortion related to the eccentric position of the receptive fields on the tangent screen. Simple cells usually had a low spontaneous activity, at least in the anesthetic state used in these experiments and compared to other cell types; they usually generated 2 to 10 spikes per minute spontaneously and only responded with a few spikes even to the optimal stimulus. This sluggish behavior made the detailed mapping of some fields difficult. The inhibitory flanks were often especially hard to demonstrate with a stationary stimulus, but their presence could usually be shown indirectly by comparing the response to a slit confined to the excitatory region with the response to a larger slit that included some or all of the inhibitory regions. In those cells in which separate "on" and "off" subareas could clearly be demonstrated, the orientation of a moving line and the most effective direction of back and forth movement could very often be predicted from the arrangement of subareas. Some cells with a side by side arrangement of two antagonistic areas responded only to a simple contrast moved in either direction. One example is the cell in figure 2B: this cell responded only when a light edge was moved from above into the on-area or when a dark edge was moved from below into the off-area - that is, it reacted to moving a light-above-dark contrast in both directions. Of cells responding specifically to edges, including simple and complex cells, the great majority preferred horizontal edges with dark below. Slits were very ineffective in activating such cells. Most cells were much more sensitive to a slit than to a dark bar; those preferring dark stimuli usually had one or two exceptionally strong inhibitory areas in their fields. For any given cell the preferred direction of movement was generally the same for a dark bar a s for a slit. The cell Fig. 3 Polar plots of responses of four complex cells. A. Cell responding best to a moving edge. B. Complex cell whose field could be subdivided, by flashing stationary stimuli, into a n upper area which gave "on" responses and a lower area giving "off' responses. T h e tuning curve of this cell was the narrowest recorded in this series. C. Cell responding best to a dark bar. D. Cell responding best to a narrow slit of light.
A
COMPLEX
CELL (235)
10 c y c l e s w i t h 5 0 % 50' square speed 30"//nac
TO'
down lighl edge
UP: dark edge
nasal c
lsmporal
' 1 ' down
Figure 3A
B COMPLEX CELL ( 2 8 5 )
/ Y
10
down
Figure 3B
MOUSE V I S U A L CORTEX
c
in figure 2C was the only exception; here a bar and a slit had diametrically opposite optimal movement directions.
COMPLEX CELL (289) IDCISI.. W)h .W.d IO.,..
